building and cost analysis of an industrial automation
TRANSCRIPT
Avrupa Bilim ve Teknoloji Dergisi
Özel Sayı 28, 1-10 , Kasım 2021
© Telif hakkı EJOSAT’a aittir
Araştırma Makalesi
www.ejosat.com ISSN:2148-2683
European Journal of Science and Technology
Special Issue 28, 1-10, November 2021
Copyright © 2021 EJOSAT
Research Article
http://dergipark.gov.tr/ejosat 1
Building and Cost Analysis of an Industrial Automation System
using Industrial Robots and PLC Integration*
Enes Efe1†, Muciz Ozcan2, Huseyin Hakli3
1* Hitit University, Department of Electrical and Electronics Engineering,Corum, Turkey, (ORCID ID 0000-0002-6136-6140), [email protected] 2 Necmettin Erbakan University, Department of EEE, Konya, Turkey, (ORCID ID 0000-0001-5277-6650) [email protected]
3 Necmettin Erbakan Univertsity, Department of Computer Engineering, Konya, Turkey, (ORCID ID 0000-0001-5019-071X) [email protected]
(1st International Conference on Applied Engineering and Natural Sciences ICAENS 2021, November 1-3, 2021)
(DOI: 10.31590/ejosat.972290)
ATIF/REFERENCE: Efe, E., Ozcan, M. & Haklı, H. (2021). Building and Cost Analysis of an Industrial Automation System using
Industrial Robots and Programmable Logic Controller Integration. European Journal of Science and Technology, (28), 1-10.
Abstract
Technology rapidly advances on a daily basis and the resulting changes can provide numerousbenefits for manufacturing methods and
machines. Manufacturers who are able to swiftly embrace thesedevelopments can increase their manufacturing output, thereby
boosting profitability and gainingcompetitive advantages over their rivals. However, the cost savings which result from new
innovationscan vary, depending on the manufacturing model. Consequently, manufacturers need to conduct accurateanalyses for
appropriate manufacturing methods in order to ensure that new changes are cost-effective.Nowadays, the use of industrial automation
systems is gaining popularity as a method of increasingprofitability for mass production, and these systems utilize control systems,
such as industrial robots andprogrammable logic controllers. The use of these elements in the manufacturing process not onlyprovides
quality and flexible production methods, which are indispensable considerations, but alsoconserves human effort. The aim of this
study was to minimize the cost of a factory-installed industrialautomation system, which produced globe valves with side couplings,
through the combined use ofindustrial robots and programmable logic controllers. While calculating returns from the installed
system,the differential evolution algorithm was used to predict future unit prices of electricity, and it wasdetermined that the cost of
investment would be recovered after a maximum of 2.5 years and that currentyearly production would increase fourfold.
Keywords: Industrial automation, Programmable logic controller, Industrial robots, Differential evolution algorithm, Prediction.
Endüstriyel Robot ve PLC Entegrasyonuyla Talaşlı İmalat Üretim
İşleminin Gerçekleştirilmesi
Öz
Üreticiler üretim şekillerini değiştirmeden önce doğru analizler yaparak kendilerine en uygun üretim yöntemini seçmeleri
gerekmektedir. Bu çalışmada yan rakorlu küresel vana üretilen bir fabrikada Endüstriyel Otomasyon Sistemi kurulmuş ve kurulan bu
sistemin maliyeti Endüstriyel Robot ve PLC beraber kullanılarak en aza indirilmesi hedeflenmiştir. Üretim yönteminde Endüstriyel
Robotun kullanılması ve mevcut sistemde ki üretimde bir takım değişiklikler yapılmasıyla esnek üretim sağlanmış ve üretimin
aksamaması için bazı tedbirler alınmıştır. Kurulan sistem maliyetinin geri dönüşüm süreci hesaplanmasında Diferansiyel Evrim
Algoritmasından yararlanılarak gelecekteki elektrik birim fiyatları tahmin edilmiştir. Bu çalışmada, yapılan yatırımın en fazla 2,5 yıl
içerisinde geri döneceği ve mevcut yıllık üretim miktarının da yaklaşık 4 kat artacağı tespit edilmiştir.
Anahtar Kelimeler: Endüstriyel Otomasyon, PLC, Endüstriyel Robot, Diferansiyel Evrim Algoritması.
* This paper was produced from the master's thesis of the first author † Corresponding Author: [email protected]
Avrupa Bilim ve Teknoloji Dergisi
e-ISSN: 2148-2683 2
1. Introduction
As technology advances apace, new innovations have a critical
effect on our lives. Manufacturing quality and the availability of
cheap products have become inevitable necessities in today’s
global markets. The use of automation technologies has made it
is possible to mass-produce items and lower production costs [1].
Automation is defined as the automatic control of a tool, process,
or system as a result of observation, decision making, and the
ability to effect changes via mechanical or electronic devices,
rather than human interaction [2]. An entire task is shared by
humans and machines, and the sharing ratio of this activity
effectively determines the level of automation. If human power
predominates, the resulting system is said to be semi-automated;
if machine power is dominant, this is known as full automation
[3]. Industrial automation (IA), which uses modern techniques
and applications in manufacturing, is an impressive
manufacturing strategy that leads to ongoing competition
between rival companies [4]. These systems are capable of
replacing human power in virtually any business sector and
utilize vital control systems, such as programmable logic
controllers (PLC), industrial robots, computers and information
technologies. This strategy enables quality and flexibility to be
increased within manufacturing operations, and, on inspecting
the manufacturing processes of developed countries, it becomes
immediately obvious when automation systems have been
widely deployed.
Industrial robots, which are among the most critical elements
of industrial automation systems, continue to grow in importance
on a daily basis [5]. Japan was the first country to use robots in
industry and, at the time, their introduction brought concerns that
unemployment would rise. However, their widening usage has
removed any doubts that this would happen and has, in fact, led
to many new lines of work, with the result that unemployment
has decreased substantially [6]. Today, they are most frequently
used in working environments where there is a risk to human life
from hazardous conditions, resulting from high temperatures,
mechanical vibrations, chemicals, and nuclear energy [7].
The PLC is another important element of industrial automation
systems. Intended to replace relay command circuits, the PLC
was so named because it was only able to perform basic logic
operations when first introduced. Firms such as Allen Bradley,
General Electric, Siemens, and Westinghouse produced the first
medium-cost and high-performance PLCs, allowing this type of
controller to be applied in industry [3]. As Toshiba, Mitsubishi,
and Omron developed low-cost, high-performance devices, their
use in industrial automation systems became even more
widespread [8]. PLCs have many attributes including, but not
limited to, flexibility, reliability, ease of expansion, and low
power consumption. It is possible to control alterations and
enlarge the system solely by changing the PLC software [9].
Various industrial automation studies have been documented
in which PLCs and industrial robots have been used. In 2007,
Niola et. al. examined the possibility of using a video system to
plan a robot’s trajectory, and this was achieved by controlling
robots with a computer mouse from a personal computer (PC)
monitor [10]. Dong and Kuang’s 2013 study analyzed the
communication signals between a Mitsubishi PLC and
GlaxoSmithKline robot [11]. In 2017, Stückelmaier et al.
presented a method of gradually increasing the trajectory
accuracy of industrial robots and examined dynamic modeling,
definition, and kinematic calibration [12]. Jeong et al. published
a study, in 2017, that discussed the software architecture of an
integrated PLC robot, which played a key role in industrial smart
systems [13]. Chen and Dong conducted studies to increase the
accuracy and efficiency of robotic engraving and examined the
feasibility of developing engraving systems that are capable of
carrying out tasks that were previously thought to be the preserve
of computer numerical control (CNC) machines [14].
In developed countries, engineering services are usually billed
on an hourly basis. However, this is not the case in developing
countries, such as Turkey, where remittance usually occurs at
monthly intervals. In addition, advanced robots are more
expensive compared to basic machines, and this results in
additional costs for developing countries. Consequently,
developed countries are able to use high-end industrial robots,
that do not require additional engineering services and are
capable of using PLC software, rather than having to procure
further control systems. However, installation costs can be
lowered by the use of a basic robot operating in conjunction with
a PLC, and a number of studies have been documented that focus
on reducing this expenditure.
This study was conducted in an active and physically-
accommodating plant and produced statistical data, which was
subsequently evaluated. In order to accurately calculate the cost
of the installed system and determine the payback period, it was
necessary to establish the price of the total electrical energy that
the system would expend. This was determined by the use of a
differential evolution (DE) algorithm to successfully predict
future unit electricity prices, thereby allowing the total electricity
bill for the system to be calculated. The DE algorithm is suitable
for a range of applications in the electrical energy sector and can,
for example, be used to determine load estimates, which are a
crucial consideration in this field. Eke successfully estimated
medium-term load using the DE algorithm [15], and Wang et al.
used the DE algorithm to predict electrical energy consumption
[16].
This study evaluated an automation system, consisting of an
industrial robot and PLC, which was installed in a factory that
manufactures side coupling globe valves (SCGV). The findings
from this exercise were assessed and the costs of the system were
compared with those that were incurred prior to installation. One
of the most important aspects of this new system was the PLC,
which, in combination with the industrial robot, enabled fast
production. In addition, this study allowed the introduction of
new techniques to calculate costs and determine the payback
period.
This paper is structured as follows: Sections 2 are concerned
with the installation of the industrial automation system and
application of the proposed method; and Section 3-4 examines
the experimental results which were obtained for the installed
system.
2. Material and Method
This study focused on the use of an industrial automation
system for the manufacture of valves with side couplings. Two
CNC machines, a 6-axis industrial robot, and a PLC were
provided for this purpose, and communications between these
elements were achieved by the use of digital input-output signals.
European Journal of Science and Technology
e-ISSN: 2148-2683 3
2.1. Industrial Robot and PLC
An industrial robot is defined, according to ISO 8373, as a
manipulator with three or more axes, which is programmable,
multipurpose, and can be either wheeled or stationary [17]. The
mobility of this manipulator varies according to its number of
axes and features, and electrical, hydraulic, or pneumatic driver
systems are used to power its joint movements. An industrial
robot essentially comprises the manipulator, a teach pendant, and
controller, while elements that remain outside the system
boundaries of the robot are termed peripherals. The end effector,
in this case, a standard pneumatic gripper, is used to grab the
working parts (Figure 1). Other peripherals include sensors,
machines, conveyor belts, and security equipment, etc. The
controller ensures that the industrial robot works in harmony
with the peripheral devices and performs the desired movements
[18].
Fig. 1 The Standard Pneumatic Gripper Used as an End
Effector in this Study.
PLCs have a microprocessor base, receive information from
detectors in the field, and process this information as instructed.
Consequently, they are able to control remote instrumentation
and can be used in industrial automation systems to execute
command and control mechanisms. In addition, PLCs are
industrial computers that are equipped with input and output
capabilities that enable them to communicate with separate
devices and work in combination with SCADA [19]. The PLC
examined in this study allowed the actions of the industrial robot
and CNC to be synchronized, thereby allowing control of the
hardware elements in the system.
2.2. DE Algorithm
The DE Algorithm was proposed by Storn and Price in 1995
and is a population-based method that uses operators resembling
those in genetic instructions, such as mutation, selection, and
crossing [20]. The DE Algorithm uses the following parameters:
population size (Np), scaling factor (F), and crossing ratio (Cr).
The principal steps of the DE Algorithm are:
Step 1: Determination of the initial population.
Step 2: Evaluation of the initial population.
Step 3: Mutation.
Mutation is a process that adds the scaled difference of two
random individuals to another individual that has been randomly
selected from the population.
𝑣𝑚,𝑡+1 = 𝑥𝑟3,𝑡 + 𝐹 × (𝑥𝑟1,𝑡 − 𝑥𝑟2,𝑡) (1)
In Equation 1, 𝑣𝑚,𝑡+1 represents the mutated individual, while
𝑥𝑟1,𝑡 , 𝑥𝑟2,𝑡, and 𝑥𝑟3,𝑡 are randomly selected individuals from the
population 𝑥𝑟1≠ 𝑥𝑟2
≠ 𝑥𝑟3≠ 𝑥𝑖).
Step 4: Crossing.
The value of 𝐶𝑟 determines which genes are taken from the
new individual, due to the mutation. If a randomly-generated
value between 0 and 1 is less than 𝐶𝑟, it is chosen from 𝑛𝑗,𝑖,𝐺+1;
if it is greater than C_r, it is selected from the current vector. The
aim of these operations is to ensure that a previously specified
ratio of genes are taken from the mutated individual. The
mathematical expression for the crossing method is:
𝑥𝑗,𝑢,𝑡+1 = {𝑥𝑗,𝑖,𝑡 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒
𝑥𝑗,𝑚,𝑡+1 𝑖𝑓 𝑟𝑎𝑛𝑑[0,1]≤𝐶𝑟 𝑜𝑟 𝑗=𝑗𝑟𝑎𝑛𝑑 (2)
In Equation 2, xu,t+1 represents the candidate individual after
crossing, and j is the current gene index. If none of the values
acquired for any gene are less than Cr, the current individual
remains unchanged. In order to circumvent this situation, a
randomly-selected gene (jrand) is updated to ensure at least one
gene mutation occurs.
Step 5: Choice.
A choice is made in favor of the candidate individual or
current individual that best fits the criteria. This practice is
generally known as the greedy-choice method. A comparison is
made between the current individual’s purpose function
performance and the candidate’s solution.
Step 6: Stopping Criterion.
Steps 3, 4, and 5 are repeated sequentially until the stopping
criterion is satisfied. Otherwise, the best solution is reported and
the algorithm terminates.
2.3. Installation of the Industrial Automation System
The components of the industrial automation system
installed during this study are shown in Figure 2. The system
comprised two CNC machines, a handling station, an industrial
robot, and PLC. The CNC machines were positioned directly
opposite each other, with the robot installed between them, and
the handling station was placed in front of the robot. The PLC
controlled the hardware elements of the system and was in
constant communication with the industrial robot. In addition to
ensuring that the industrial robot and the CNCs worked in
synchronization, the PLC controlled the operation of the
handling station, while the industrial robot commanded the CNC
machines. A program was written and uploaded to the PLC, in
order to execute the necessary control operations.
Fig. 2 Illustration of the Control Components of the System.
Two grippers were used to exchange processed and
unprocessed parts. The first gripper carried the completed SCGV
Avrupa Bilim ve Teknoloji Dergisi
e-ISSN: 2148-2683 4
away from the CNC, while the second gripper transported the
unprocessed part to the CNC (Figure 3).
(a) Unprocessed Part.
(b) Processed Part.
Fig. 3
Having selected appropriate grippers, it was necessary to
design a handling station that was suitable for use with the robot
arm. Processed parts were carried to the conveyor by the robot.
The exchange of parts between the handling station and
industrial robot was made possible by the use of optical
proximity sensors, which were keyed at a frequency of 1000 Hz.
When the part had been completed, the sensors sent a ready
signal to the robot, which seized it using the gripper. Figure 4
depicts the position of the sensors on the handling station, while
Figure 5 shows the use of the gripper to seize a part.
Figure 4. Handling Station
Sensors.
Figure 5. Gripper
Receiving a Part.
In order to ensure that the actions of the handling station and
CNC machines were coordinated, it was necessary to use a PLC
with a sufficient number of inputs and outputs. The PLC was
required to control the handling station, CNC gates, and mirrors,
and the manufacturer’s proprietary ladder programming language
was used to ensure this functionality was satisfied. In addition to
being operated via the industrial robot, the CNC machine gates
and mirrors were also controlled by the PLC. Manual operation
was therefore possible when the industrial robot was switched
off. When the system starts up, the industrial robot initially takes
the unprocessed part from the handling station (Figure 6 (a)) and
then hands it to the CNC for processing, while simultaneously
taking the completed part away from the machine (Figure 6 (b) -
(c)). Lastly, the robot places the processed part on the conveyor.
(a)
(b)
(c)
(d)
Fig. 6 Industrial Robot Actions.
Control of the CNC machine’s gates was switched from
manual to automatic using dual pneumatic valves, a 40 x 40 x
600 mm piston, and magnetic sensors, which operated at a
switching frequency of 5000 Hz (Figure 7).
(a)
(b)
Fig. 7 (a) Piston. (b) Dual Pneumatic Valve.
Finally the industrial robot was programmed with its integrated
programming language, which resembled C, to ensure system
stability.
2.4. Electricity Price Prediction Using the DE Algorithm
In order to observe the system’s efficiency and determine the
payback period, it was necessary to estimate the unit price of
electricity in the following years. Since electricity prices depend
on a multitude of parameters, they can be highly variable over a
period of time. It was therefore necessary to adopt the use of
robust and reliable methods, through the use of smart systems, to
provide suitable forecasts. Although there are no documented
studies concerning the calculation of unit electricity prices,
numerous enquiries have been carried out into the estimation of
future energy demand. Many nature-inspired algorithms have
been used for this purpose and these include a number of
computational methods, such as the genetic algorithm [21, 22],
particle swarm optimization [23, 24], ant colony optimization
[25], artificial bee colony optimization [26], the DE algorithm
[27], and various hybrid techniques [28]. Studies have also been
undertaken with regard to the estimation of electricity generation
and consumption [29-31]. This study adopted the use of the DE
algorithm, which, in addition to offering efficient performance
and being easy to apply, was based on the work of Beskirli et al.
[27], who proposed its use for energy demand estimation in
preference to other nature-inspired methods.
Four economic criteria were considered when estimating
energy demand: gross domestic product (GDP), population, and
values of imports and exports. [23-27]. These parameters,
together with electricity generation and consumption, directly
influence unit electricity prices. Table 1 contains values for these
6 criteria between 1995 and 2015 [29].
European Journal of Science and Technology
e-ISSN: 2148-2683 5
Table 1. Unit electricity price (kWh), GDP, population, values of imports and exports, electricity production and electricity
consumption for Turkey between 1995 and 2015.
Name and number of variables
1 2 3 4 5 6
Year Unit Price
(kWh)
GDP
($10^9)
Population
(10^6)
Import
($10^9)
Export
($10^9)
Electricity
production
(10^9)
Electricity
consumption
(10^9)
1995 0.20 168.08 58.48 35.71 21.64 86.25 67.39
1996 0.36 181.07 59.42 43.63 23.22 94.86 74.16
1997 0.71 188.73 60.37 48.56 26.26 103.30 81.89
1998 1.14 270.94 61.32 45.92 26.97 111.02 87.71
1999 1.94 247.54 62.28 40.67 26.59 116.44 91.20
2000 3.43 265.38 63.24 54.50 27.78 124.92 98.30
2001 4.72 196.73 64.19 41.40 31.33 122.73 97.07
2002 10.62 230.49 65.14 51.55 36.06 129.40 102.95
2003 12.00 304.90 66.08 69.34 47.25 140.58 111.77
2004 11.22 390.38 67.00 97.54 63.17 150.70 121.14
2005 11.22 481.49 67.90 116.77 73.48 161.96 130.26
2006 10.00 526.42 68.76 139.58 85.54 176.30 143.07
2007 9.39 648.75 69.59 170.06 107.27 191.56 155.14
2008 14.37 742.09 70.44 201.96 132.03 198.42 161.95
2009 14.15 616.70 71.33 140.93 102.14 194.81 156.89
2010 15.31 731.60 72.32 185.54 113.88 211.21 172.05
2011 15.31 773.97 73.40 240.84 134.91 229.40 186.10
2012 15.81 786.28 74.56 236.54 152.47 239.50 194.92
2013 18.16 823.04 75.78 251.66 151.81 240.15 198.05
2014 18.16 799.36 77.03 242.17 157.62 251.96 207.38
2015 19.28 719.62 78.27 207.23 143.84 261.78 217.31
Examination of Table 1 indicates that, although electricity
generation and consumption have steadily increased over the
years, there are periods of time when electricity unit prices have
stayed relatively constant, or even slightly declined, prior to
subsequent increases. In general, GDP and the values of imports
have risen steadily between 1995 and 2013, although they have
fallen in the last two years of the period considered.
Two separate models, one linear, the other quadratic, were
implemented to estimate energy or electricity demand [23, 25-
27]. Of these, the quadratic model proved to be much more
efficient and was preferred for this study. The quadratic equation
for four variables can be expressed as follows:
𝐸𝑞𝑢𝑎𝑑𝑟𝑎𝑡𝑖𝑐 = 𝑤1 + 𝑤2𝑋1 + 𝑤3𝑋2 + 𝑤4𝑋3 + 𝑤5𝑋4 +
𝑤6𝑋1𝑋2 + 𝑤7𝑋1𝑋3 + 𝑤8𝑋1𝑋4 + 𝑤9𝑋2𝑋3 + 𝑤10𝑋2𝑋4 +
𝑤11𝑋3𝑋4 + 𝑤12𝑋12 + 𝑤13𝑋2
2 + 𝑤14𝑋32 + 𝑤15𝑋4
2 (3)
The parameters 𝑋1, 𝑋2, 𝑋3, 𝑋4 represent GDP, population,
values of imports, and values of exports, respectively, while 𝑤1,
𝑤15 are their corresponding weights. Equation 3 is formulated
for 4 variables, and if this number increases, the formula must be
changed accordingly. For 4 variables, there are 15 weights (also
the dimensions of the problem); when there are 5 variables there
are 21 weights; and 6 variables require 28 weights. The accuracy
of the prediction is verified by the purpose function shown in
Equation 4.
𝑚𝑖𝑛𝑓(𝑣) = ∑ (𝐸𝑟𝑜𝑏𝑠𝑒𝑟𝑣𝑒𝑑𝑅
𝑟=1 − 𝐸𝑟𝑝𝑟𝑒𝑑𝑖𝑐𝑡𝑒𝑑
)2 (4)
𝐸𝑟𝑜𝑏𝑠𝑒𝑟𝑣𝑒𝑑 and 𝐸𝑟
𝑝𝑟𝑒𝑑𝑖𝑐𝑡𝑒𝑑 are the observed and predicted unit
electricity prices, respectively, and R is the year in which the
observation occurs. The main objective is to find the weights
which correspond to the 𝑚𝑖𝑛𝑓(𝑣) value for each year [27]. The
smaller 𝑓(𝑣) is, the more accurate the estimate for a given year
is.
3. Results
In this study, experimental testing was carried out using a PC
equipped with an i5-6400 central processing unit (CPU), an
AMD Radeon R7-200 graphics processing unit (GPU), and 8
gigabytes (GB) of random-access memory (RAM). The
Industrial Automation System was realized in firm facilities. The
experimental results were separated into three parts; an
Avrupa Bilim ve Teknoloji Dergisi
e-ISSN: 2148-2683 6
evaluation of the utility of the system, unit electricity price
estimates, and cost analysis and payback period calculations.
3.1. Analysis of the Industrial Automation System Output
A significant increase in production was noted after the
industrial robot and PLC had been installed. Prior to automation,
the daily production output, between 8 a.m. and 6 p.m., was an
average of 400 units, while after installation the output had risen
to an average quantity of 585 units. If the system was allowed to
run round the clock, this output jumped to around 1400 units.
Previously, daily production was limited to approximately 8
hours, after the workers’ lunch breaks and recess times had been
taken into account. Following automation, continuous production
was possible on a 24-hour basis, and this was the most
significant benefit to result from installation of the system.
The system provides an additional advantage in that the PLC
can allow manual production, if required. Machining is only
possible with the use of the industrial robot; however, manual
production is only possible if the PLC and robot are active.
3.2. Future Projections of Unit Electricity Price
The data presented in Table 1, over a 21 year period, was used
for the estimation of electricity unit prices, using the DE
algorithm. In this analysis, F was taken to be 0.5, while Cr and
Np were set to values of 0.9 and 100, respectively. The
maximum number of function evaluations was specified as
5x105. The variables GDP, population, and values of imports
and exports were initially tested as stand-alone criteria;
electricity generation and consumption were then tested in
combination with the other parameters (Table 2). The quadratic
form given in Equation 3 was rearranged in accordance with the
number of variables used.
The DE algorithm was run using variables that were chosen for
4 different scenarios, and, after 10 trials, the best results were
selected for use in this study. Table 2 gives the weights and
errors for the 4 variables used.
Table 2. Comparisons of Coefficients and Relative Errors.
Selected Variables
Coefficients 1,2,3,4 1,2,3,4,5 1,2,3,4,6 1,2,3,4,5,6
w1 10.00000 -9.17532 9.62941 -8.48483
w2 -0.24023 0.09704 3.32920 -2.61183
w3 -1.29403 -9.99305 1.60644 9.02448
w4 -3.00539 2.19926 -4.12544 -8.70575
w5 6.02331 -4.09795 -6.21791 -1.76599
w6 0.00453 3.31954 -6.13156 9.94205
w7 0.00122 -0.00483 -0.08804 -7.33863
w8 0.00204 -0.01348 -0.00783 -0.01581
w9 0.04775 0.00536 -0.02306 -0.03232
w10 -0.09946 -0.00096 0.02342 0.21898
w11 -0.00991 -0.03412 0.15556 0.49497
w12 -0.00041 0.14561 0.20481 -0.54486
w13 0.01937 -0.15470 0.04591 -0.29004
w14 -0.00037 0.06002 0.04389 0.43933
w15 0.00681 0.01262 -0.07044 1.65158
w16 - -0.05923 -0.05802 -1.34516
w17 - 0.00168 0.00228 0.58050
w18 - 0.24909 0.01065 1.91664
w19 - -0.00033 0.01433 -2.07848
w20 - -0.03886 0.03802 -2.73528
w21 - 0.03174 0.01783 2.86035
w22 - - - -1.90207
w23 - - - -0.01904
w24 - - - -0.55196
w25 - - - -0.17802
w26 - - - -0.85209
w27 - - - -0.38560
w28 - - - 2.64912
Error Rate 33.21 9.79 21.83 3016.10
Examination of Table 2 reveals that the best electric unit price
estimation, with a 9.79 error rate, resulted from the combination
of variables 1, 2, 3, 4, and 5. When electricity generation and
consumption were both included in the calculations, the error
rate for the estimate became significantly higher in relation to
other conditions. Incremental increases in the error value
corresponded to higher fitting values, which are the product of a
square function [27]. Future projections relied on the
combination of GDP, population, values of imports, and values
of exports.
Figure 8 compares the estimates obtained for variables 1, 2, 3,
4, and 5, with the values observed between 1995 and 2015. The
results show that, in general, the estimations were very close to
the actual values, and it can be concluded that the proposed
model was reliable and fit for purpose. This is particularly true
when examining the findings for 2000, 2007, and 2009; although
some disparity is evident for 2002 and 2003.
European Journal of Science and Technology
e-ISSN: 2148-2683 7
Fig. 8
Comparison of Observed and Estimated Electricity Unit Prices.
Three different scenarios were prepared for the estimation of
electricity unit prices in Turkey, between 2016 and 2035.
Scenario 1: Between 2016-2035, the following were assumed:
a mean GDP growth rate of 3%, a population growth rate of 1%,
an import growth rate of 3%, and a 5% increase in electricity
production and values of exports.
Scenario 2: Between 2016-2035, the following were assumed:
a mean GDP growth rate of 4%, a population growth rate of 2%,
an import growth rate of 4%, and a 5% increase in electricity
production and values of exports.
Scenario 3: Between 2016-2035, the following were assumed:
a mean GDP growth rate of 6%, a population growth rate of 4%,
an import growth rate of 6%, and a 7% increase in electricity
production and values of exports. Table 3 contains future
projections that were obtained for each scenario, using the
weight values acquired from the DE algorithm. These scenarios
were determined by considering low, normal, and high levels of
increase in the appropriate variables. Examination of the findings
reveals that Scenario 1 resulted in the highest level of agreement
between the estimations and actual figures, from 2016 to 2018.
The rapid increase in unit price observed for Scenario 2 and less
inflationary trends of Scenarios 1 and 3 are illustrated in Figure
9. It can be seen that after 2022, the estimates obtained for
Scenario 3 were initially smaller than those of the other data sets;
after 2030, however, they quickly increased and went on to
exceed the values derived for Scenario 1. The results show that
the estimated figures for Scenarios 2 and 3 continued to increase
over time, while the values obtained for Scenario 1 fell over the
last two years of the time frame.
Fig. 9 Future Projections of Electricity Unit Price for Scenarios 1-3.
0
2
4
6
8
10
12
14
16
18
20
19
95
19
96
19
97
19
98
19
99
20
00
20
01
20
02
20
03
20
04
20
05
20
06
20
07
20
08
20
09
20
10
20
11
20
12
20
13
20
14
20
15
Ele
ctri
city
Un
it P
rice
Years
Observed
Estimated
1030507090
110130150170190210230250270290310330350
2016 2018 2020 2022 2024 2026 2028 2030 2032 2034
Scenario 1 Scenario 2
Avrupa Bilim ve Teknoloji Dergisi
e-ISSN: 2148-2683 8
3.3. Systems Costs and Profit Analysis
Future projections of the electricity unit price were used to
carry out a cost and profit analysis, and the following tariffs were
added to these values: a 40% distribution charge for unilateral
industry groups, a 1% energy fund, a 5% municipal consumption
tax (MCT), and 18% value added tax (VAT). Table 3 shows the
resulting estimates for all three scenarios.
Table 3. Electricity Prices Estimates Including Tax (Turkish
kurus).
Scenarios
Year 1 2 3
2019 39.36 64.54 105.85
2020 42.94 70.42 115.49
2021 47.23 77.46 127.04
2022 52.23 85.66 140.48
2023 57.89 94.93 155.69
2024 64.14 105.19 172.52
2025 70.89 116.27 190.68
2026 78.00 127.93 209.80
2027 85.28 139.86 229.37
2028 92.48 151.67 248.73
2029 99.28 162.83 267.04
2030 105.30 172.69 283.21
2031 110.04 180.46 295.95
2032 112.89 185.14 303.64
2033 113.14 185.56 304.31
2034 109.92 180.26 295.63
2035 102.16 167.54 274.76
Two different modes of operation were examined during this
study. In the first of these (Mode a), it was assumed that the new
system would work continuously for 24 hours; in the second
mode (Mode b), it was assumed that work would only take place
between the hours of 8 a.m. and 6 p.m. The total energy costs of
the system were calculated for both operating modes, using the
previous cost estimates with taxes (Table 4). In addition, the
ticket prices of the control elements of the system were included,
and the average power consumption of these components was
taken as 5.5 kW and multiplied by the electricity prices including
tax. The total number of working days was assumed to be 260,
allowing for weekends, and the results were multiplied by this
figure in order to calculate the annual energy costs of the system.
Table 4. Total Energy Costs of the System for Various
Scenarios and Operating Modes (103 Turkish lira).
Scenarios/ Operating Modes
1 2 3
Year a b a b a b
2019 13.51 5.63 22.15 9.23 36.33 15.14
2020 14.74 6.14 24.17 10.07 39.63 16.51
2021 16.21 6.75 26.59 11.08 43.60 18.17
2022 17.93 7.47 29.40 12.25 48.21 20.09
2023 19.87 8.28 32.58 13.58 53.43 22.26
2024 22.01 9.17 36.10 15.04 59.21 24.67
2025 24.33 10.14 39.90 16.63 65.44 27.27
2026 26.77 11.15 43.90 18.29 72.00 30.00
2027 29.27 12.20 48.00 20.00 78.72 32.80
2028 31.74 13.22 52.05 21.69 85.37 35.57
2029 34.07 14.20 55.88 23.28 91.65 38.19
2030 36.14 15.06 59.27 24.69 97.20 40.50
2031 37.76 15.74 61.93 25.81 101.57 42.32
2032 38.74 16.14 63.54 26.48 104.21 43.42
2033 38.83 16.18 63.68 26.53 104.44 43.52
2034 37.72 15.72 61.87 25.78 101.46 42.27
2035 35.06 14.61 57.50 23.96 94.30 39.29
It was assumed that workers would be paid the minimum
wage. Since accurate forecasts for the inflation rate were critical
for reliable estimates, appropriate data was obtained from the
Central Bank of Turkey. After installation, labor costs were
largely replaced by the energy costs of the system, and the
difference between them was multiplied by 260 to give the
profits shown.
The price increases for SCGVs were determined in accordance
with the inflation rate forecasts. SCGV unit prices were
multiplied by the average daily production figure of 400 and
number of working days (260) to obtain the annual income prior
to installation. The corresponding revenue after installation was
calculated in a similar fashion, and daily production estimates of
1400 and 585 units were assumed for Modes a and b,
respectively.
Table 5 presents the cumulative total profit for each scenario
and operating mode over the period in question. It should be
noted that the manufacturer paid an initial system installation
cost of 42,000 euros.
European Journal of Science and Technology
e-ISSN: 2148-2683 9
Table 5. Cumulative Total Profit After Installation of the System (103 Turkish lira).
Scenarios
1 2 3
Years a b a b a b
2019 403.40 93.43 394.75 89.82 380.57 83.92
2020 847.03 196.02 828.96 188.49 799.31 176.14
2021 1334.62 308.47 1306.17 296.61 1259.51 277.17
2022 1870.22 431.47 1830.30 414.83 1764.83 387.55
2023 2458.33 565.80 2405.70 543.87 2319.37 507.90
2024 3104.08 712.50 3037.36 684.69 2927.93 639.10
2025 3813.16 872.67 3730.86 838.38 3595.89 782.14
2026 4591.86 1047.59 4492.43 1006.16 4329.36 938.22
2027 5447.20 1238.67 5329.04 1189.43 5135.25 1108.69
2028 6386.96 1447.47 6248.49 1389.77 6021.39 1295.15
2029 7419.85 1675.81 7259.57 1609.03 6996.71 1499.50
2030 8555.56 1925.74 8372.15 1849.32 8071.35 1723.99
2031 9804.89 2199.54 9597.31 2113.05 9256.87 1971.20
2032 11179.85 2499.80 10947.47 2402.98 10566.37 2244.19
2033 12693.86 2829.43 12436.62 2722.25 12014.77 2546.47
2034 14361.85 3191.69 14080.47 3074.45 13619.02 2882.17
2035 16200.49 3590.27 15896.68 3463.68 15398.43 3256.08 Initial installation cost was 42,000 euros (222,306 TL, according to exchange rate of 01.06.18).
Figure 10 shows the profit graphs for the scenarios and
operating modes in question. It can be seen that if the automated
system commenced operation in 2019, it would become
profitable within 2-3 years, assuming 8 a.m. to 6 p.m. working.
However, if the system was operated on a 24 hour basis, it would
become profitable within the first year. Although theoretically
possible, producers prefer to avoid continuous production due to
the attendant reduction in machine service life.
Fig. 10 Profit Graph of the System.
4. Discussion and Conclusion
This study considered the installation of an industrial
automation system which produced SCGVs. One of the
innovations that were done in this study is that the Industrial
Automation System computes the time to return to profitability
by using artificial intelligence techniques. In contrast to many
other systems, a PLC was used in addition to an industrial robot
and, in order to accurately calculate the costs of the installed
system to the producer, a DE algorithm was used to estimate
future electricity unit prices in Turkey. Thus the total energy cost
was calculated and compared with the labor costs prior to
installation. In the light of this study, the system was expected to
become profitable within 2.5 years and return considerable gains
thereafter.
The use of a PLC provides significant advantages, namely: the
ability of the system to operate manually, providing continuous
production, and the option to procure cheaper robots without
additional features. Furthermore, the resulting simplicity of the
robot software increases the usability of the system. In summary,
the installation of this automated system resulted in increased
production and higher levels of efficiency and flexibility.
In the future, it may be possible to achieve greater efficiency
by reducing energy expenditure. The electric energy used by the
system could be supplied from renewable energy sources, and
the system could be made more economical. Moreover, the use
of a wireless communication feature could increase the usability
of the system by enabling control over a remote computer.
References
1. Yücel, İ.H., Sanayide Robot Teknolojisi. Uygulaması Ve
Önemi, DPT Yayınları, Ankara, Aralık, 1991.
2. Sheridan, T.B., Telerobotics, automation, and human
supervisory control. 1992: MIT press.
3. Özcan, M., Otomasyon sistemlerinde PLC uygulamaları.
2004: Atlas Yayın dağıtım.
Avrupa Bilim ve Teknoloji Dergisi
e-ISSN: 2148-2683 10
4. Acharya, V., S.K. Sharma, and S. Kumar Gupta, Analyzing
the factors in industrial automation using analytic hierarchy
process. Computers & Electrical Engineering, 2017.
5. Çengelci, B. and H. Çimen, Endüstriyel robotlar. Makine
Teknolojileri Elektronik Dergisi, 2005. 2(2): p. 69-78.
6. Peşkircioğlu, N., Otomasyon ve Entegre Kalite Kontrolu.
Verimlilik Dergisi, (15): p. 19-40.
7. De Silva, D., Reactions to Robots, Engineering. 1987, April.
8. Kurtulan, S., PLC ile endüstriyel otomasyon. 2001: Birsen
Yayınevi.
9. Rehg, J.A. and G.J. Sartori, Programmable logic controllers.
2009, Upper Saddle River, N.J.: Pearson Prentice Hall.
10. Niola, V., C. Rossi, and S. Savino, Vision system for
industrial robots path planning. International journal of
Mechanics and control, 2007. 8(1): p. 35-45.
11. CHEN, D.-q. and Y.-c. KUANG, Communication between
Mitsubishi PLC and GSK Industrial Robot. Mechanical
Engineer, 2013. 4: p. 023.
12. Stückelmaier, P., M. Grotjahn, and C. Fräger. Iterative
improvement of path accuracy of industrial robots using
external measurements. in Advanced Intelligent
Mechatronics (AIM), 2017 IEEE International Conference
on. 2017. IEEE.
13. Jeong, H.S., et al. Design of SW architecture for PLC
integrated robot. in Ubiquitous Robots and Ambient
Intelligence (URAI), 2017 14th International Conference on.
2017. IEEE.
14. Chen, Y. and F. Dong, Robot machining: recent
development and future research issues. The International
Journal of Advanced Manufacturing Technology, 2013.
66(9-12): p. 1489-1497.
15. Eke, İ., Diferansiyel evrim algoritması destekli yapay sinir
ağı ile orta dönem yük tahmini. Uluslararası Mühendislik
Araştırma ve Geliştirme Dergisi, 2011. 3(1): p. 28-32.
16. Wang, L., et al., Effective electricity energy consumption
forecasting using echo state network improved by
differential evolution algorithm. Energy, 2018. 153: p. 801-
815.
17. Standard, I., 8373: 1994. Manipulating industrial robots–
Vocabulary.
18. Dişlitaş, S., Endüstriyel Robot Programlama. Baskı,
Endüstriyel Robot Programlama Eğitimi ile Mesleki ve
Teknik Eğitim Güçlendirilmesi (ERPE-METEG) Projesi,
Çorum, 2015.
19. Mirzaoğlu, İ. and M. Sarıtaş, PLC VE SCADA Kullanarak
Bir İrmik Üretim Sisteminin Otomasyonu. 2008.
20. Storn, R. and K. Price, Differential evolution-A simple and
efficient adaptive scheme for global optimization over
continuous spaces [R]. Berkeley: ICSI, 1995.
21. Canyurt, O.E., et al., Energy demand estimation based on
two-different genetic algorithm approaches. Energy Sources,
2004. 26(14): p. 1313-1320.
22. Ceylan, H. and H.K. Ozturk, Estimating energy demand of
Turkey based on economic indicators using genetic
algorithm approach. Energy Conversion and Management,
2004. 45(15-16): p. 2525-2537.
23. Unler, A., Improvement of energy demand forecasts using
swarm intelligence: The case of Turkey with projections to
2025. Energy Policy, 2008. 36(6): p. 1937-1944.
24. Gulcu, S. and H. Kodaz, The estimation of the electricity
energy demand using particle swarm optimization
algorithm: A case study of Turkey. 8th International
Conference on Advances in Information Technology, 2017.
111: p. 64-70.
25. Toksari, M.D., Estimating the net electricity energy
generation and demand using the ant colony optimization
approach: Case of Turkey. Energy Policy, 2009. 37(3): p.
1181-1187.
26. Sonmez, M., A.P. Akgungor, and S. Bektas, Estimating
transportation energy demand in Turkey using the artificial
bee colony algorithm. Energy, 2017. 122: p. 301-310.
27. Beskirli, M., H. Hakli, and H. Kodaz, The energy demand
estimation for Turkey using differential evolution algorithm.
Sadhana-Academy Proceedings in Engineering Sciences,
2017. 42(10): p. 1705-1715.
28. Kiran, M.S., et al., A novel hybrid approach based on
Particle Swarm Optimization and Ant Colony Algorithm to
forecast energy demand of Turkey. Energy Conversion and
Management, 2012. 53(1): p. 75-83.
29. Yumurtaci, Z. and E. Asmaz, Electric energy demand of
Turkey for the year 2050. Energy Sources, 2004. 26(12): p.
1157-1164.
30. Kavaklioglu, K., et al., Modeling and prediction of Turkey's
electricity consumption using Artificial Neural Networks.
Energy Conversion and Management, 2009. 50(11): p. 2719-
2727.
31. Toksari, M.D., A hybrid algorithm of Ant Colony
Optimization (ACO) and Iterated Local Search (ILS) for
estimating electricity domestic consumption: Case of Turkey.
International Journal of Electrical Power & Energy Systems,
2016. 78: p. 776-782.